Crop type maps are essential for informing, assessing, and managing agricultural practices and food security. Traditionally, ground truth data for these maps is collected through field surveys, which are time-consuming, labor-intensive, and difficult to scale. As a result, large-scale crop type mapping faces challenges in obtaining comprehensive and high-quality reference data. Street view imagery and vehicle-based surveys offer alternative solutions for field inspections, providing scalable and cost-effective ways to collect crop type information.
This 2-hour workshop will enable participants to: 1. Explore the potential of street view imagery as a novel and emerging source for large-scale crop type ground truth data collection. 2. Understand CropSight, the GeoAI-driven workflow for retrieving object-based crop type ground truth information, including collecting geotagged street view images, extracting crop type labels from street view images, and delineating crop field boundaries corresponding to each retrieved label using satellite imagery.
3. Discuss key challenges, best practices, and opportunities for integrating street view imagery into remote sensing workflows.
Participants of all skill levels in Python are welcome, and we will use Google Colab to ensure accessibility and ease of use. To support hands-on learning, we will provide example datasets and Jupyter notebooks for the workshop.
Authors Zhijie Zhou (zhijiez2@illinois.edu), Tianci Guo (tiancig2@illinois.edu), Yin Liu (yinl3@illinois.edu), Chunyuan Diao (chunyuan@illinois.edu) Department of Geography and GIScience, University of Illinois Urbana-Champaign
